The Synergy Register: Tracking Promised Value to Delivery

A synergy register turns one promised number into tracked lines. Each line names the change, the person accountable for it, the baseline it is measured against, the amount, the cost of obtaining it, the date it lands and the evidence that it arrived. Fix no baseline before completion and no line can ever be shown as delivered or missed, and the document becomes a list of intentions.

Underneath that answer sits one uncomfortable fact about measurement. A saving is not a number anybody can walk into a warehouse and count. A saving is a difference between what a business actually spent and what it would otherwise have spent, and the second half of that comparison never happens. Once two businesses are running as one, the version in which they stayed separate has stopped existing. So the only way to keep it is to write it down first. Everything a register does, every field it carries and every argument about how it is kept, comes back to protecting that written-down comparison from the pressure that builds against it later.

What is a synergy register, and who keeps it?

A synergy register is a single document in which every promised element of value from a purchase has its own row and its own history. Not a slide. Not a monthly pack. One document, one keeper, one row per promise, and a record of what that row said at every point since it was created.

Think about how a household tracks a decision to switch from two scooters to one car. The claim is that the household will spend less. Six months later nobody will remember what the two scooters cost to run, and everybody will have an opinion. So somebody writes that figure down over the last year, before the car arrives. If that number is written down in advance and kept, the household can settle the argument. If it is not, the argument runs forever and the person who feels most strongly wins it. A synergy register is that sheet of paper, scaled up to a purchase worth crores. Several people now have an interest in what the sheet says.

The keeper matters more than the format. One named person maintains the document, and that person is deliberately not any of the people accountable for delivering the individual rows. A register is not a report: a report is produced for a meeting and a register is maintained continuously, and that difference is what decides whether anybody can trust the document six months later. A meeting asks for the current state, so a report shows the current state. A register shows the current state and every state before it. The interesting question is almost never what a row says today, but when the row started saying that, and what it said before.

At Harivansh Packaging Limited the natural keeper is Devyani Kulkarni, its chief financial officer. Ashwin Rege led the transaction team, so the opening content comes most naturally from him. The two roles are separate on purpose. The person who built the argument for the purchase should not be the person who marks it delivered. The separation is not an accusation about anybody. The same person cannot be both the advocate and the scorer without the score drifting, and the drift is not deliberate enough for anyone to notice it happening.

What does a single line in the register contain?

Eight fields. Six of them can be filled in on the day the promise is made, in a room with the plan on the table. Two of them cannot, and those two are exactly the ones that decide whether the row is a measurement or a wish.

The two fields most often missing are the cost of obtaining and the evidence, and a register without them records ambition rather than delivery. Notice what those two have in common. Both of them make the row smaller or harder. The cost of obtaining reduces the number in the amount column. The evidence commits somebody in advance to a test they might fail. Every other field can be filled in with enthusiasm intact. The cost of obtaining and the evidence cannot, and they go missing for exactly that reason. A register that carries both is worth several times one that does not.

Why does the baseline have to be fixed before completion?

The baseline carries the argument the whole document rests on. A rule this easy to nod at is just as easy to violate quietly, so the argument is worth taking slowly.

A saving is a difference between two quantities. The first is what the combined business actually spent. The amount actually spent is observable: it appears in the accounts, somebody can look it up, and two people will get the same answer. The second is what the two businesses together would have spent if the purchase had never happened. The would-have-spent figure is not observable at all, and it never becomes observable. A figure of that kind is a counterfactualEvents that did not happen, kept alongside the events that did, so a change has something to be measured against., and the moment the two businesses start operating as one it stops being available to anybody.

So the baseline is not a measurement, it is a decision recorded in advance, and the only thing that makes it useful later is that it was recorded before anybody knew what they would want it to say. Two sensible people, working from the same figures before completionThe day a purchase legally takes effect and the buyer takes control. Signing comes earlier, and the two are often months apart., can set a baseline differently and both be defensible. One takes the last full year. Another takes the last four quarters adjusted for a plant that was shut. Neither is wrong. Setting the baseline afterwards is not defensible, and there the disagreement is not about method at all.

Here is why setting it afterwards is a different kind of mistake rather than a slightly worse version of the same one. Suppose the baseline is drawn six months after completion, from whatever the management accountsThe monthly or quarterly numbers a business produces for itself. Faster and less formal than the accounts it publishes, and not audited. showed at that point. By then some of the promised changes have already partly happened. Suppliers have been renegotiated. Two sites have started running as one. The numbers in those accounts therefore already contain part of the result the baseline is supposed to measure against. Every row now compares itself to a position that includes some of its own effect, and the size of that contamination is unknown and unknowable.

The practical instruction that falls out of all this is short enough to remember. The baseline is a document, dated before completion, signed by somebody, and never edited afterwards. If a correction is genuinely needed, it is recorded as a correction with its own date and reason, sitting beside the original rather than replacing it. The moment a baseline can be quietly overwritten, every argument about delivery becomes an argument about what the baseline used to say, and nobody can win that argument because the evidence has been deleted.

How to build an M&A Synergy Register

Seven steps, in this order, and the order carries as much of the work as the steps do. In a different sequence, the work runs into baselines being set for rows that should have been deleted, or people being named against rows whose evidence nobody has agreed.

Step two usually removes more rows than anybody expects, and that is the register working rather than the register failing. Ask of each row a single question: did this change need the purchase to happen? A great many of the things that land on a first draft did not. Renegotiating a freight contract, closing a duplicate software licence, tightening a shift pattern. All worth doing. None of them needed anybody to pay anybody anything. Leaving them in does not make the business better off by one rupee; it only makes the register look fuller, and it guarantees that when somebody later asks whether the purchase paid for itself, the answer will be padded with things that would have happened anyway.

The other steps are less dramatic and each removes one specific later argument. Step three removes the argument about what the comparison was. Step four removes the argument about whose job it was, the argument that consumes the most time and produces the least. Step five removes the argument about whether a number is gross or net. Step six removes the argument about whether something counts as delivered. Step seven removes the argument about when to look. Fixing a review date sounds trivial until a date has slid by three weeks four times in a row, each slip excused by a difficult quarter.

What total does this register have to add up to?

Now put the document against the transaction it belongs to. Harivansh Packaging Limited bought Sundarban Polymers Private Limited. The enterprise value was Rs 1,320 crore; strip out Sundarban Polymers' net debt of Rs 180 crore and Rs 1,140 crore is what actually reached the sellers. The buyer found that money in two places: Rs 140 crore it already held as cash, and a fresh facility of Rs 1,000 crore carrying its own contracted rate of 9.0 per cent.

The accretion arithmetic on that purchase runs as follows. Harivansh Packaging earned Rs 225 crore of profit after tax. Bring in Sundarban Polymers at Rs 61 crore, a rounded value, then charge the interest on the fresh facility, Rs 67.5 crore once the 25.0 per cent effective tax rateThe tax charge in a set of accounts divided by the profit before tax. The effective rate is the rate the accounts actually show, rather than any rate written in a statute. has done its work. The three together come to Rs 218.5 crore. Nothing has happened to the share count, and 18.00 crore shares still divide the profit, so the figure per share drops to Rs 12.14/- from the Rs 12.50/- it was. The purchase is dilutiveDescribes a purchase where profit rose by less than the cost of paying for the deal, so each existing share is entitled to less profit than before. on day one.

Getting back to Rs 12.50/- means pushing profit after tax up to Rs 225 crore again, and the distance is Rs 6.5 crore. Because a quarter of every additional rupee goes in tax, that Rs 6.5 crore has to come from Rs 8.67 crore of further earnings before interest, tax, depreciation and amortisation, or EBITDAA rough measure of what the trading operation earns in a year, taken before the cost of borrowing, the tax bill and the writing down of machinery come off it., counted after whatever was spent to obtain it. Rs 8.67 crore is the total this register has to add up to: set beside the Rs 132 crore of EBITDA Sundarban Polymers earns, that is 6.6 per cent, and against the Rs 609 crore that the combined business earns, 1.42 per cent.

The record fixes no figure for what the transaction team promised in this purchase. The Rs 8.67 crore above was not promised by anybody. The figure was derived from the accretion test, as the amount that would restore the position per share. A register in a real programme is built against a promise. The register built here is worked against a requirement.

How much does one rupee of delivery actually move?

Once the total is fixed, every individual line becomes priceable, and the arithmetic is small enough to do mentally. A quarter of every rupee goes in tax, so each rupee of delivered EBITDA carries 75 paise into profit after tax. Dividing by 18.00 crore shares gives the per share effect. Worked at a convenient size: every Rs 1.80 crore of delivered EBITDA moves earnings per share by about Rs 0.075/-, so any line can be priced the moment its size is known.

Now the same arithmetic in the form a review meeting actually uses. Deliver half the requirement, Rs 4.33 crore, and earnings per share reaches about Rs 12.32/-, still Rs 0.18/- below the Rs 12.50/- the buyer started from. A register reporting half its total delivered has not restored the position. Fifty per cent delivered gets read in meetings as though the problem is half solved, and in the only sense that matters to a shareholder it is not solved at all. The position per share is still lower than it was before the purchase. Half of the way back is not back.

How is delivery evidenced, and what counts as evidence?

Two documents can say the identical thing and mean completely different things, and this is where most registers quietly stop being useful.

Evidence is a figure somebody outside the programme could reproduce from the accounts, and a status field marked delivered by the person accountable for delivering it is not evidence but a claim. Neither remark is a comment on anybody's honesty. The difference is in what the two objects are. One is a number with a trail behind it, and a stranger following that trail arrives at the same number. The other is an opinion held by a person with an interest, and no trail exists for anybody to follow.

The evidence standard is agreed at the same time as the baseline. Left until afterwards, it gets negotiated rather than agreed. The reason is entirely predictable. Before completion, nobody knows which rows will go well, so everybody argues about the standard in the abstract and lands somewhere reasonable. Eighteen months later, everybody knows exactly which rows are in trouble, and the argument about what counts as evidence is really an argument about those specific rows, conducted in the language of principle. Setting the standard early is not a matter of rigour. The point is to fix the standard while nobody yet has a reason to want a particular answer.

A workable standard looks different for each kind of row. For a purchasing saving, it is a figure derived from a supplier ledger that the finance team can pull without asking the programme for anything. For a change expressed as a run rateThe yearly figure a monthly number would come to if this one month simply repeated itself twelve times over., it is the same number computed from three consecutive months rather than the best one. For anything that shows up as headcount, it is the payroll rather than an organisation chart. A chart records a decision. A payroll records what happened.

Why does the cost of obtaining belong on the same row?

Take the illustrative line just described. Rs 3 crore of annual benefit, Rs 2 crore to obtain it. The line contributes Rs 1 crore. The subtraction is not a subtle point, and yet the presentation choice that hides it is extremely common: put Rs 3 crore in the register's amount column and put the Rs 2 crore in a separate programme cost budget, held by a different person, reported in a different pack, on a different cycle.

A register showing Rs 3 crore in the amount column while the Rs 2 crore of cost sits in another document is overstating that line by three times, and nothing in either document is untrue. The truthfulness is what makes the presentation dangerous. Every number is right. The formatting is what lies. And because the two documents have different owners and different cycles, the two figures are almost never in the same room at the same time, so the overstatement never gets noticed by anybody who could correct it.

How do lines quietly disappear, and how is that prevented?

Four ways, and here is the thing worth noticing about all four: none of them requires anybody to be dishonest. Each one is a reasonable-sounding administrative act performed by a busy person, and the effect is that a promise leaves the document without any decision ever being taken to remove it.

Each in turn. Merging is the most innocent looking. Two rows describe changes at the same site, somebody sensibly reports them together, and from that quarter onwards there is one row where there were two. Nothing was hidden. But the merged row now has one baseline where there were two, one date, one status, and if it misses, nobody can say which half missed.

Restating in a different unit is subtler. A row created as an annual rupee amount starts being reported as a percentage improvement, or as tonnes, or as a headcount. Each of those may be a better operational measure. None of them can be compared with the baseline the row was created against, so the row is now unmeasurable while still appearing on the document.

The moving date is the one everybody has seen. The amount never changes, the row is never challenged, and the delivery date advances by one quarter at every review. After four reviews the row has been on the register for a year, has delivered nothing, and has never once been reported as behind. At each review it was reported against a date that had just been reset.

The fourth is the reclassification. A row created as a saving, meaning a cost the business used to pay and now does not, gets restated as a cost avoidanceMoney a business says it did not have to spend, as against money it has stopped spending. Nothing in the accounts goes down when a cost is avoided., meaning a cost the business says it would have incurred and did not. The difference is that the first shows up in the accounts and the second does not. A register that lets rows drift in that direction can report a large total while the combined figures show nothing at all.

None of the four requires anybody to be dishonest, and all four become visible the moment the register keeps its history rather than only its current state. Keeping the history is the whole defence, and it is why the register-against-report distinction is not a matter of housekeeping. Look at any row and ask what it said four quarters ago. If the document can answer, all four failures are catchable. If it cannot, none of them is.

What happens to a line that will not be delivered?

The line gets written off, formally, with a date and a reason, and the total comes down.

A write-off feels like an admission of failure and is treated as one in most programmes, and for exactly that reason it rarely happens. But look at what the two options actually produce. A write-off is information and a line carried at zero progress for two years is noise, so the formal write-off is the better result.

Follow the consequence of each branch. Written off, the total falls by the row's amount, the reason is on the record, and everybody planning against that total now has a true number to plan against. The lender who lent Rs 1,000 crore into this purchase, the board, the person building next year's budget, all of them now know something they did not know. Carried at zero, the total stays where it was, and it stays there for two years describing a change that everybody involved privately knows will not happen. The document has the authority of a document, so it is now a worse guide than none at all.

A register showing no write-offs after a year is not a register that is going well. Some rows always fail. The world moves, a supplier is taken over, a system replacement runs late, a key person leaves. If a document covering twenty or thirty promised changes has not recorded a single miss after twelve months, the plausible explanation is not that everything worked, but that nobody has yet been willing to be the person who records a miss.

Who reads the register, and what decision does it support?

Three audiences, in this order of priority, and getting the order wrong is what quietly ruins the document.

First, the people who have to deliver the rows. For them the register is a working list: what is mine, what am I measured against, what evidence will be asked for, when. If it does not serve them, it will be filled in badly. A document that only exists to be presented gets treated as a presentation task.

Second, the board. For them the register answers one narrow question: is the argument for this purchase holding up? Not whether the purchase was a good idea, a different question that no document settles. Just whether the specific changes the price was justified by are arriving, and on what evidence.

Third, whoever writes the review afterwards. The reviewer arrives two years later with no memory of any of it, and everything they can conclude depends on whether the history survived. If the register kept only its current state, that reader can describe the present and nothing else.

A register designed primarily to be presented becomes optimistic, and the fix is structural rather than cultural: keep the history rather than only the current state. People cannot be exhorted into candour on a document whose only visible artefact is a green status. When the previous states stay visible, candour becomes the path of least resistance, and a row that quietly improved its own story is obvious to anybody scrolling across it.